Re: Functions of Classifiers (in a conlang)

> The ambiguity between the two isn't especially odious, so long as
> you're going for a naturalistic lang.  This is relatively minor
> compared to some other sorts of ambiguity we find in natlangs.
>
Really? Do you know an example of a language which uses classifiers for
all three functions?

> Of course, if one of your design goals is the minimalization of
> ambiguity, then here are the beginnings of two possible solutions:
>
> - Make the classifier precede its NP (as above) and use a verb-initial
> constituent order.
>
> - Make the classifier follow its NP and use a verb-final constituent
> order.
I don't think this solves things if classifiers also function as pronouns:

VERB INITIAL ORDER

see I CLS.MAN walk go

I saw the walker and I/he went

or:

I saw him (the man) and walked off

Whatever word order you choose, there are ambiguities of this sort if
classifiers have both pronominal and relativizing functions.

>
> If you wish to try this -- using special forms of the classifiers as
> pronouns -- but are concerned about making these derived pronouns too
> long, consider non-concatenative morphology.  For example, the
> historical pronominalizing suffix could have been -/i/, but now the
> pronominalization of classifiers is realized by vowel umlaut, or
> consonant mutation, etc.
Yes, this is a possibility. :) I could go with the good old spoken word
solution too: stress classifiers used as pronouns and have them be
unstressed clitics when they're classifying something explicitly mentioned.

Yahya wrote:

>
>Malayo-Polynesian languages (not "East Asian", I think)
>make extensive use of "numeral classifiers", AKA
>"numeral coefficicents", also using them as anaphors.
>
>
>
Well, many East Asian languages do have numeral classifiers. I meant to
include the Sino-Tibetan languages, Japanese etc as well as the
Malayo-Polynesian languages.

>Is that where she says (in your quote):
> "(v) Noun classifiers are often used anaphorically; they may
> grammaticalize as markers of syntactic functions" ?
>
>
>
More specifically, the preceding bit: "noun classifiers are often used
anaphorically". She also goes into more detail later in the chapter though.


>Let's simplify this example first, to the essential
>conflict:
>
> MAN.CLS man come and MAN.CLS shout
>
>Is this:
>A previously-unmentioned man came and he shouted
>or...
>Another man came and the previously-mentioned man shouted
>?
>
>This example leads us to focus on the sequence of
>reference.  Do you want the same pronoun for the
>previously-mentioned as for the newly-mentioned
>man?  If not, how do we differentiate them?  I note
>that, for example, Ilomi has pronouns for this/that/
>the other, allowing three distinct referents of the
>same grammatical type, eg person or thing.  IIRC, this
>is also true of several natlangs (but can't cite details
>ATM).
>
>
>
This is not the only conflict though, although it might seem so at
first. :) Remember that I was proposing for nominality to be dependent
on classifiers... assuming that is true, then we could have things like
the following:

su? - burn
pa: - CLS.wood

pa: su?

= PRONOUN BURN = it (wood) burns

= CLASSIFIER BURN = that which burns and is wood = fire wood

My point is that the issue is not only back reference vs new reference
if stems are ambiguous between nominal and verbal interpretations, but
also truly that of whether stems are intended to be interpreted as
arguments or verbs themselves.



>Let's try instead -
>
>ii)a) Build pronoun-sequencers on classifiers.
>For example, some ending could be added to
>the classifiers to mark how far back the
>reference to this an entity classified using
>this classifier was.  Eg:
>
> MAN.CLS = "he"
> MAN.CLS-PASTNESS.COUNT = "he, previously mentioned"
>
>More specifically, if MAN.CLS = "le", and if
>the first three cardinal numbers are 1 = "sa",
>2 = "du" and 3 ='"ti", then:
>
> "he, just mentioned" = "le"
> "he, mentioned before 'le'" = "le-sa"
> "he, mentioned before 'le-sa'" = "le-du"
> "he, mentioned before 'le-du'" = "le-ti"
>
>I doubt you'd ever want to refer back to a
>fourth man, but if you did, this pattern would
>be clearly and uniquely extensible.
>
>You could also reverse this pattern:  eg English
>permits "the first-mentioned", "the second-
>mentioned", and so on.
>
>There may be some overlap between these
>pronouns and demonstratives.  Eg "this" and
>"that" may connote "the near one", "the far
>one", and even "the one upriver", "the one
>downriver" or "the one in the bush" ...  Where
>reference by location is possible, reference
>by counting is less attractive.
>
>
>
This is actually a very interesting idea... and I don't mind a little
additional morphology so much if it provides interesting information.
I'm not sure whether I agree with three degrees of reference though (I'd
prefer only 2: recent/new mention vs distant mention): I think some form
of deictic marking like "this", "that", "the other" that you mentioned
is probably better.

Reply